Liquid Crystal Phase Transitions in Systems of Colloidal Platelets with

Sep 17, 2009 - We have studied a system of polydisperse, charged colloidal gibbsite platelets with a bimodal distribution in the particle aspect ratio...
0 downloads 10 Views 2MB Size
13476

J. Phys. Chem. B 2009, 113, 13476–13484

ARTICLES Liquid Crystal Phase Transitions in Systems of Colloidal Platelets with Bimodal Shape Distribution A. A. Verhoeff,*,† H. H. Wensink,‡ M. Vis,† G. Jackson,‡ and H. N. W. Lekkerkerker† Van’t Hoff Laboratory for Physical and Colloid Chemistry, Debye Institute, Utrecht UniVersity, Padualaan 8, 3584 CH Utrecht, The Netherlands, and Department of Chemical Engineering, Imperial College London, South Kensington Campus, London SW7 2AZ, United Kingdom ReceiVed: March 30, 2009; ReVised Manuscript ReceiVed: July 17, 2009

We have studied a system of polydisperse, charged colloidal gibbsite platelets with a bimodal distribution in the particle aspect ratio. We observe a density inversion of the coexisting isotropic and nematic phases as well as a three-phase equilibrium involving a lower density nematic phase, an isotropic phase of intermediate density, and a higher density columnar phase. To relate these phenomena to the bimodality of the shape distribution, we have calculated the liquid crystal phase behavior of binary mixtures of thick and thin hard platelets for various thickness ratios. The predictions are based on the Onsager-Parsons theory for the isotropic-nematic (I-N) transition combined with a modified Lennard-Jones-Devonshire cell theory for the columnar (C) state. For sufficiently large thickness ratios, the phase diagram features an I-N density inversion and triphasic I-N-C equilibrium, in agreement with experiment. The density inversion can be attributed to a marked shape fractionation among the coexisting phases with the thick species accumulating in the isotropic phase. At high concentrations, the theory predicts a coexistence between two columnar phases with distinctly different concentrations. In experiment, however, the demixing transition is pre-empted by a transition to a kinetically arrested, glassy state with structural features resembling a columnar phase. I. Introduction The phase behavior of bidisperse colloidal systems incorporates many intriguing phenomena such as the formation of stable superstructures and phase equilibria involving multiple phases. For example, mixtures of spheres with a diameter ratio of around 0.5 are known to form AB2 and AB13 binary crystals, as first observed in natural opals1,2 and reproduced in suspensions of latex particles3,4 and colloidal hard spheres.5-7 Mixtures of long and short rod-like colloids exhibit a demixing of the nematic phase which leads to a coexistence of two nematic phases with different compositions including a triphasic isotropic-nematicnematic equilibrium, as observed experimentally8,9 and confirmed by theory.10 A similar behavior has been observed recently in binary mixtures of thin and thick hard rods with equal length,11 as originally predicted with an Onsager-Parsons description.12 In contrast to rods, the phase behavior of plate-shaped colloids has only been explored over the past decade.13-18 A remarkable feature of some of these systems is that the formation of a partially crystalline columnar phase is not suppressed by the considerable spread in particle size,15 as is the case with crystalline and smectic order in polydisperse systems of spheres19 and rods.20 Moreover, a large spread in the plate thickness may give rise to a so-called isotropic-nematic density inVersion, in which the isotropic phase becomes denser than * To whom correspondence should be addressed. E-mail: a.a.verhoeff@ uu.nl. † Utrecht University. ‡ Imperial College London.

the coexisting nematic. This phenomenon originates from a pronounced fractionation in the thickness of the particles among the coexisting phases with the thick species accumulating in the isotropic phase and the thin ones in the nematic phase. Although the nematic phase has a higher particle concentration than the isotropic phase, its mass density falls below that of the isotropic phase due to a high fraction of thin platelets with a lower particle mass. This phenomenon has been observed in a polydisperse system of sterically stabilized gibbsite platelets21 and analyzed with a simple theory based on Onsager’s secondvirial theory extended to binary mixtures.22 A similar density inversion can be predicted from fundamental measure theory based on the Zwanzig model for binary hard platelets.23 In this Article, we report our findings on a system of charged gibbsite platelets, which forms a remarkable three-phase equilibrium, involving a low-density nematic phase, an isotropic phase of intermediate density, and a dense columnar phase. A size analysis using atomic force microscopy (AFM) reveals that the distribution of the aspect ratio, defined as the plate thicknessto-diameter ratio, is characterized by a distinct bimodal shape. In an analogous fashion to the sterically stabilized system,21 the charged plate system exhibits a density inversion of the isotropic and nematic phases induced by a strong fractionation effect with respect to the aspect ratio. The fractionation scenario is corroborated by the size distributions measured in the coexisting phases, which demonstrate that the isotropic phase is rich in thick platelets, while the corresponding nematic phase consists mainly of thin platelets corresponding to a small aspect ratio.

10.1021/jp902858k CCC: $40.75  2009 American Chemical Society Published on Web 09/17/2009

Liquid Crystal Phase Transitions

J. Phys. Chem. B, Vol. 113, No. 41, 2009 13477

Figure 1. (a) Transmission electron micrograph and (b) atomic force micrograph of the gibbsite platelets. (c) Histograms of the diameter, thickness, and aspect ratio D/L of the platelets, determined with atomic force microscopy.

A theoretical underpinning of the observed effects is provided by calculating the phase diagram of a binary mixture of thin and thick platelets for various thickness ratios. This is done by combining the Onsager-Parsons theory for the isotropic to nematic transition with a modified Lennard-Jones-Devonshire cell theory for the columnar state.24 For a sufficiently large ratio of the particle thickness, the I-N density inversion and triphasic isotropic-nematic-columnar coexistence can both be satisfactorily reproduced. Moreover, at high particle concentrations, our theory predicts a demixing of the columnar phase into a fraction containing predominantly thin species and a “mixed” fraction with equal portions of thin and thick platelets. This demixing could not be observed in the experimental system due to the formation of a dynamically arrested glassy state at high particle concentrations. II. Materials and Methods Gibbsite platelets were synthesized by hydrothermal treatment of aluminum alkoxides,25 and subsequently treated with aluminum chlorohydrate to increase particle stability. The enhanced stability is associated with the presence of aluminum polycations which are formed upon addition of the aluminum salt. At pH ≈ 4, the cations hydrolyze and adsorb on the surface of the gibbsite particles. This gives rise to a steep repulsive particle interaction which prevents the platelets from aggregating.26 The gibbsite platelets were analyzed with transmission electron microscopy (TEM) using a Tecnai 10 microscope (FEI Company) and atomic force microscopy (AFM) utilizing a multimode scanning probe microscope (Digital Instruments). AFM measurements were performed in tapping mode with a standard TESP silicon tip. Samples were prepared by dilution of the gibbsite dispersion with a 1:1 water-ethanol mixture. The diluted suspension was subsequently spread onto carboncoated copper grids (TEM measurements) or freshly cleaved mica (AFM measurements). To obtain the distributions of the particle dimensions, the diameter, thickness, and aspect ratio of 200-270 particles were measured. The gibbsite platelets had a well-defined hexagonal shape (see the micrographs in Figure 1a and b), with an average diameter of 207 nm with a standard deviation in the average σD of 35% and an average thickness of 8.2 nm with a σL of 46%. The diameter of the platelets had a rather broad, symmetrical distribution, while the thickness was strongly peaked with a long tail toward the thicker particles (see Figure 1c). The aspect ratio, however, was bimodal, with a relatively narrow distribution of thick species and a much wider spread of thin ones.

The gibbsite dispersion was concentrated via centrifugation and redispersion with 10-2 M NaCl to the concentration where phase separation was just hindered by kinetic arrest. Samples of decreasing concentration were prepared upon dilution and stored in 50 × 4 × 0.2 mm3 sized glass capillaries (VitroCom), which were flame-sealed and subsequently glued to avoid evaporation of the solvent. The suspensions were left to phase separate and equilibrate at room temperature for 2 weeks. The phase behavior of the system was studied on a macroscopic scale by placing the capillaries between crossed polarizers, where they were photographed with a Nikon Coolpix 995 digital camera. The samples were examined in more detail using a Nikon LV100 Pol polarizing microscope with 10× and 20× Nikon ELWD Plan Fluor objectives, and equipped with a QImaging MicroPublisher 5 megapixel CCD camera. Bragg reflections were observed with the polarizing microscope while the sample was illuminated with a cold light source (Dolan Jenner, model 190). III. Phase Behavior Experiments In Figure 2a, we show a concentration series of the charge stabilized gibbsite platelets. Upon increasing the particle concentration, the system enters the isotropic-nematic biphasic regime, with a nematic phase at the bottom and an isotropic phase at the top of the capillary, separated by a sharp interface. Further increasing the particle concentration leads to the onset of a density inversion, where the isotropic phase starts to become denser than the nematic phase. Around the inversion point, the I-N interface shows a regular pattern of fingers reminiscent of a Rayleigh-Taylor instability (see the second capillary of Figure 2a and the enlarged view in part b).27 At higher concentrations, the inverted state is reached with a nematic top phase and an isotropic bottom phase, separated by a sharp interface. Upon subsequent densification, the system enters a triphasic region, where the isotropic phase is located between a nematic upper phase and a columnar bottom phase, all separated by sharp interfaces. The columnar bottom phase is also characterized by bright Bragg reflections (see Figure 2c), which originate from the two-dimensional hexagonal arrangement of the columnar stacks of platelets. The color associated with the Bragg reflections ranges from green to red, which corresponds to an average intercolumnar distance of about ∼250 nm. When the particle concentration is further increased, the isotropic phase disappears, leaving an upper nematic phase in coexistence with a lower columnar phase. Finally, at very high concentrations, an arrested state is reached which no longer undergoes

13478

J. Phys. Chem. B, Vol. 113, No. 41, 2009

Verhoeff et al.

Figure 2. (a) Concentration series of suspensions of charged gibbsite platelets. A sequence of phase equilibria is observed, ranging from biphasic isotropic-nematic (I-N), inverted biphasic isotropic-nematic (I-N), triphasic isotropic-nematic-columnar (I-N-C), biphasic nematic-columnar (N-C), to a columnar glass (CG). (b) Regular fingering pattern in an isotropic-nematic sample at a density of 20 v/v %, close to the inversion point. Bragg reflections in the columnar phase (c) and in the dynamically arrested columnar glass (d).

macroscopic phase separation. Birefringent patterns are clearly visible and persist over a long time. These correspond to “frozenin” regions of particle alignment caused by the shear forces during the filling of the capillary. The sample also shows Bragg reflections originating from small grains with a domain size of about 5 µm, as depicted in Figure 2d. This indicates that, despite the high packing fraction, the platelets still have enough free volume to locally self-assemble into columnar arrangements. The capillary containing the inverted isotropic-nematic phase equilibrium was further investigated by splitting the capillary and analyzing the two coexisting phases separately with AFM. In Figure 3, we depict the sample between crossed polarizers together with AFM images of the gibbsite platelets in both phases. The measured diameter, thickness, and (inverse) aspect ratio distributions of the platelets in the isotropic and nematic phases are presented in Figure 4. Although the diameter distribution is virtually the same in both phases, the thickness distributions are significantly different, with the isotropic phase markedly enriched in thick platelets. The most distinct difference can be inferred from the histograms showing the aspect ratio distribution in both phases. From this, it is evident that the nematic phase is characterized by a unimodal distribution of thin platelets (large D/L), whereas the isotropic phase possesses a bimodal shape distribution with a significant fraction of thick species (small D/L). IV. Onsager-Parsons Theory for the Isotropic-Nematic Transition A simple theory is now employed to describe the main features of the experimental system. For the present system of charged platelets, a theory would be required based on a hardcore model supplemented with a suitable electrostatic potential. Unfortunately, the effective (screened) pair potential between two uniformly charged platelets resulting from the Poisson-

Figure 3. (a) Image of a gibbsite suspension with isotropic-nematic density inversion, taken between crossed polarizers. AFM images of gibbsite platelets sampled from the nematic (b) and isotropic (c) phase.

Boltzmann theory is not known in analytical form28 and multipole expansionssaimed at providing an analytical approximation valid at large plate separationssare questionable at high particle densities (in particular in the nematic and columnar phases). Headway can be made by restricting the plate

Liquid Crystal Phase Transitions

J. Phys. Chem. B, Vol. 113, No. 41, 2009 13479

orientations to the three Cartesian axes. Within the so-called Zwanzig approximation, a tractable density functional theory can be formulated which allows one to scrutinize the effect of charge on the structure and phase behavior of colloidal platelets.29 Alternatively, interaction site models, based on a discretization of the surface charge into Yukawa sites compacted onto a circular disk-shaped array, can be analyzed with integral equation theory to elucidate the pair correlation functions of charged platelets at low to moderate densities.30 The scope of our current paper is to keep the theory as tractable as possible while maintaining the essential physics of the problem. Therefore, we will build on the simple but physically plausible idea that the main effect of the double layers is to alter the effective shape of the platelets, such that the effective thickness by which the platelets interact is enhanced. The use of an effective hard particle is merited by the fact that the electric double layers surrounding the platelets are rather tight.42 Let us consider a binary mixture of N ) N1 + N2 hard cylinders with two different lengths, L1 and L2, and common diameter D ) D1 ) D2 in a macroscopic volume V. For the plate-like cylinders we consider here, the inverse aspect ratio Li/D (i ) 1,2) is much smaller than unity. Henceforth, we shall assign the label “2” to the thickest particles. The overall concentration of the system is expressed in dimensionless form via c ) ND3/V. Following ref 22, the Onsager-Parsons Helmholtz free energy of such a mixture at a given c and mole fraction xi ) Ni/N is given by

βF ˜ - 1) + ) (ln Vc N

∑ xi{ln xi + 〈ln 4πfi〉i} +

i)1,2

cGP(φ) 2

ij (γ)〉〉ij ∑ ∑ xixj〈〈V˜excl i

(1)

j

where β-1 ) kBT is the thermal energy and V˜ ) ∏i V xi i/D3, with Vi being the thermal volume of species i (including contributions arising from the rotational momenta of the platelet). The brackets 〈( · )〉i ) ∫dΩfi(Ω)( · ) denote an orienta-

tional average according to some unknown orientational distribution function fi, normalized via ∫dΩfi(Ω) ) 1. Several entropic contributions can be distinguished in eq 1. The first three are exact and denote the ideal translational, mixing, and orientational entropy, respectively. The last term represents the excess translational or packing entropy which accounts for the particle correlations on the approximate second-virial level by considering pair interactions only. The key quantity here is the ij between two plate-like cylinders of type excluded volume Vexcl i and j at fixed interparticle angle γ:31

(

ij Vexcl (γ)

π ) |sin γ| + 2 D3 LiLj Li Lj π π + + E(sin γ) + cos γ + 2 2 |sin γ| D D 4 4 D

ij V˜excl (γ) )

)(

)

(2)

with E(x) being the complete elliptic integral of the second kind. Although the structure of eq 1 is similar to the classic Onsager second-virial free energy, higher order virial terms are incorporated approximately through a density rescaling according to Parsons’ recipe32-34 as extended to mixtures (see ref 35 for a recent review of the use of the Onsager-Parsons free energy to describe mixtures of hard particles). This involves a rescaled density cGP where the factor GP ) (1 - 3φ/4)/(1 - φ)2 depends on the total plate volume fraction φ ) c∑i xi(πLi/4D). Note that GP approaches unity in the low-density limit φ f 0 in which case the original second-virial theory is recovered, as required. One must then specify the orientational averaging. By definition, all orientations are equally probable in the isotropic (I) phase and fi ) 1/4π. The orientational contributions then simply become

〈ln 4πfi〉i ≡ 0

(I)

(3)

Using the isotropic averages 〈〈sin γ〉〉 ) π/4, 〈〈E(sin γ)〉〉 ) π2/8 and 〈〈cos γ〉〉 ) 1/2, we obtain for the excluded volume contribution

Figure 4. Histograms of the platelet diameter (a), thickness (b), and aspect ratio (c) sampled from the nematic and isotropic phase of a sample with isotropic-nematic density inversion (see Figure 3), determined with AFM.

13480

J. Phys. Chem. B, Vol. 113, No. 41, 2009

(

)(

Verhoeff et al.

)

LiLj π Li Lj 3π π2 π2 ij 〈〈V˜excl (γ)〉〉 ) + 2 + + + 8 D D 8 8 D 2

(4) In the nematic (N) phase, the particles are expected to be oriented along a common nematic director. For strongly ordered states, it is expedient to adopt a Gaussian trial function to describe the orientational probability density.36 For a uniaxial nematic phase, the Gaussian trial function takes on the following form:

fi(θ) ≡

{

Ri 1 exp - Riθ2 4π 2

[ ] R 1 exp[- R (π - θ) ] 4π 2 i

2

i

π if 0 e θ e 2 π